spec-driven-development
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a gated development workflow involving human review at every phase (Specify, Plan, Tasks, Implement). This process forces manual validation of requirements and technical plans before any code is generated or executed, providing a strong procedural security control.
- [SAFE]: The repository includes extensive security documentation in
references/security-checklist.md, covering threat modeling, input validation, secure authentication, and supply chain hygiene. It specifically advocates for blocking and reviewing dependency lifecycle scripts. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user requirements to generate specifications and tasks. While it has capabilities to write files and suggest build commands, the risk is mitigated by the mandatory human review gates and the practice of reframing vague requirements into concrete, testable success criteria.
- Ingestion points: User feature requests and clarifying questions (Phases 0 and 1 in
SKILL.md). - Boundary markers: Uses structured Markdown templates for specifications and capability maps to delimit AI output.
- Capability inventory: File creation and modification (specs, task lists, plans); execution of standard development tools (
npm,git,npx). - Sanitization: Relies on human review at every phase transition and explicit 'Assumption' surfacing to prevent unintended instruction following.
Audit Metadata